»ö« Welcome to Perl 6! | perl6.org/ | evalbot usage: 'perl6: say 3;' or rakudo:, niecza:, std:, or /msg p6eval perl6: ... | irclog: irc.perl6.org/ | UTF-8 is our friend!
Set by sorear on 4 February 2011.
00:05 Transformer joined 00:06 Transformer left 00:34 woosley joined 00:47 Alias_ joined 00:49 Alias left 00:50 kensanata left 01:03 am0c joined 01:34 [Coke] left 01:35 Bucciarati left 01:36 [Coke] joined 01:42 Chillance left 01:50 [Coke] left 01:52 [Coke] joined 02:05 [Coke] left 02:06 [Coke] joined 02:07 Bucciarati joined 02:11 Bucciarati left 02:16 [Coke] left 02:18 [Coke] joined 02:19 sftp left, sftp joined 02:25 [Coke] left 02:26 Bucciarati joined 02:27 [Coke] joined 02:31 estrabd left, [Coke] left 02:33 [Coke] joined, estrabd joined 02:34 nymacro joined 02:37 Bucciarati left 02:46 Bucciarati joined 02:47 alester joined 02:56 Bucciarati left 03:00 Bucciarati joined 03:01 skyheights joined
skyheights was happy to see all the activity on #perl6 yesterday 03:02
03:02 skyheights left
TimToady too 03:06
03:06 JimmyZ left 03:12 Juerd left 03:13 PerlJam left, Util left, Util joined, Juerd joined, hatseflats left 03:14 PerlJam joined, hatseflats joined 03:18 JimmyZ joined 03:29 am0c left 03:32 hudnix left, [Coke] left 03:34 [Coke] joined 03:48 [Coke] left 03:49 araujo left 03:50 envi joined, [Coke] joined 03:56 orafu left, orafu joined 04:02 [Coke] left 04:03 JimmyZ left, [Coke] joined 04:13 [Coke] left 04:15 [Coke] joined 04:22 [Coke] left 04:24 [Coke] joined 04:44 alester left
TiMBuS www.winestockwebdesign.com/Essays/L...Curse.html i dont get it 05:34
doing things in lisp is so easy that nothing gets done ? 05:35
TiMBuS looks at cpan
so, perl is.. hard?
06:04 gente joined, gente left 06:08 woosley left 06:09 woosley joined
s1n TiMBuS: that guy is a goon, proof: 06:15
"Endgame: A random old-time Lisp hacker's collection of macros will add up to an undocumented, unportable, bug-ridden implementation of 80% of Haskell because Lisp is more powerful than Haskell."
06:19 woosley left 06:40 envi left 06:44 envi joined 06:56 araujo joined
dalek rixel: d7e7c61 | diakopter++ | / (11 files):
some cursor work and a bit of non-basic-block-detecting optimization of simple variables that aren't captured, making them CIL locals. meh.
 07:00
07:01 justatheory left
sorear diakopter: interesting-sounding 07:01
diakopter ish 07:02
07:03 dju__ joined
diakopter it needs a full basic-block pass or two 07:03
perlesque: my $x = 2_000_000_000; 1 while --$x; # shouldn't time out (any longer; it used to) 07:05
p6eval perlesque: ( no output )
diakopter my desktop pc is a lot faster than p6eval 07:06
07:06 dju_ left
diakopter sorear: i guess it's fully effective if all the computation heavy basic blocks were manually separated out by the programmer 07:09
well, sorta
out& 07:10
07:20 Tedd1 left 07:26 Tedd1 joined 07:30 _jaldhar left 07:31 _jaldhar joined 07:45 daxim joined 07:55 molaf joined 07:56 donri joined 08:36 plobsing left 08:41 plobsing joined
ncow ok, I understand what people have said earlier about the long compile times for rakudo, I really do. The machine in question is a PIII with 256 mb ram, so no exactly top of the line. It can usually build Perl 5 from source, for instance, in around 30 min. So far, building rakudo for the first time, it's been over 30 hours. Honestly, this is getting ridiculous. I really can't remember anything taking so long to compile on this machine. I could have built ha 08:42
I'm not trying to critize rakudo or perl 6, but anything that takes more then a day to compile is difficult, imho, to want to invest a lot of time into, despite the fact that I want to finally check this out (and I still want to, assuming it ever finishes compiling) 08:43
08:44 Axius joined
frettled ncow: you need to load splitlong.pl, your first statement stopped at «I could have built ha» 08:45
ncow I'm trying really hard, and I've built pretty much everything there is from source under Linux, and I really can't recall anything taking more than 30 min to build (other than Linux 2.6 kernel, which usually takes 3-4 hours max to finish building both the kernel and modules)
frettled: I'm not sure I understand? Are my lines coming as truncated to you? I'm fairly certain my lines are maxing out at a normal length? 08:47
frettled In 1994, building X took more than 24 hours on a SparcStation 2, which was closer to state-of-the-art than your now ancient hardware. When Mozilla (the web browser-mail client-news client) arrived on the scene, the source was around 1 GB, and also took more than 24 hours to compile.
ncow: Yes, they do come as truncated, if they're too long. Use splitlong.pl.
ncow frettled: are you sure? the hardware I'm building on was new in 2003, nearly 10 years /later/
so how could my hardware be ancient compared to 1994 hardware 08:48
frettled ncow: it's ancient _today_. 08:49
ncow: building these programs I spoke of took ages on more-current hardware than your hardware is current today
ncow frettled: I honestly think that it's your irc client that is truncated. I can clearly see my lines coming from the server on my other client and it looks fine
frettled ncow: while that hardware was merely 1-3 years old at the time, yours is 8 years old.
ncow: No, it's not my IRC client. It's how IRC works. Trust me on this.
ncow: Load splitlong.pl. Guess why it's even needed... 08:50
ncow frettled: why do I see the full likes in /another/ client then
frettled That your client prints what you type is just how it is, and always have been.
ncow: it was only the ONE statement that was truncated, the others were not.
moritz ncow: irclog.perlgeek.de/perl6/2011-04-17#i_3526308 08:51
ncow I'm not talking about /this/ client's local echoing, I'm talking about another client (on my phone) that I can see all the lines (including the one you meantioned) in full, which is why I believe it's your client doing that, as it's clearly not the server doing it
frettled ncow: The IRC protocol does not support more than 512 8-bit characters on a single line.
ncow frettled: not every server strictly adheres to the RFCs 08:52
s/servers/daemon software/
frettled ncow: /script load splitlong.pl
just do it
moritz ncow: back to the original topic... what do you actually want?
frettled ncow: Yes, and sending lines more than 512 chars is in violation of the protocol, which is why you must split those lines.
huf so it doesnt even get cut off at the same length for all servers on freenode? _nice_ 08:53
ncow and where exactly am I supposed to get this script from? You seem to be assuming you even know what client I'm on right now
frettled ncow: Yes, I did a /ctcp ncow version, and got the response that you're using irssi.
moritz ncow: we know that the compilation time is less than awesome, and there's work under way to fix it
ncow huf: apparently
huf good work irc, just like when you specified the encoding of the messages properly... 08:54
ncow "/script load splitlong.pl" gives me unknown command
huf just dont write long lines then
or do, and expect them to be cut off and play this "fun" game every time
ncow I usually try not to, and to be honest, I didn't think my lines were all that long
huf this is why we load splitlong ;)
ncow I get that. I'm still wondering why I can see the full lines on my phone that I sent from my laptop 08:55
oh well
I'm off to bed, sorry for any trouble
frettled ncow: okay, you've probably not installed scriptassist, then. Please follow the instructions here instead: bugs.launchpad.net/ubuntu/+source/...comments/3
ncow ok, noted 08:56
frettled ncow: don't fret about it :)
tadzik ncow: pardon me, but didn't you complain about exactly the same thing yesterday? 09:00
moritz tadzik: which is why I asked what exactly he wanted. I didn't get any reply 09:01
tadzik I think that question was already answered
moritz seems that line-length-in-IRC discussions are more interesting 09:02
frettled Anyway, I think it's hard to solve long compile times on eight year old hardware with a low amount of RAM - 256 MB isn't much, it wasn't much eight years ago, either (1024 seemed pretty common then), so it's not unreasonable to expect trashing because of low memory. However, as moritz++ noted, Rakudo does take a rather long time to compile even on more current hardware with plenty of RAM, even though it's just a matter of tens of minutes rather ... 09:03
... than tens of hours.
(w00t, splitlong.pl in action)
sbp hitting enter more often works too :-) 09:04
frettled Hmm, I haven't tried compiling Rakudo on my new-ish Mac Mini, which has an SSD.
sbp: that looks too much like work ;)
moritz doesn't think rakudo compilation is CPU bound, except when swapping :-)
frettled spends a lot of time and energy being lazy. Works well with being a sysadmin - laziness (automation) is _good_.
sbp could be a good slogan for perl6! 09:05
moritz .u ellipsis
phenny U+0EAF LAO ELLIPSIS (ຯ)
sbp "Automated laziness."
moritz .u horizontal ellipsis
phenny U+2026 HORIZONTAL ELLIPSIS (…)
moritz something to teach splitlong.pl to use :-)
frettled perl Configure.pl --gen-parrot for Rakudo Star 2011.01: Around 2.5 minutes on a shared-time computer with a 2.4 GHz CPU and 8 GiB RAM.
frettled is now running make -j8 :) 09:06
moritz: not compatible with US-ASCII and Latin-1, but perhaps splitlong.pl could check irssi's configuration regarding UTF-8-ness. 09:07
huf ascii and latin1 can go where the sun dont shine
frettled sbp: nice slogan :)
tadzik isn't jnthn's work on lexical-settings going to help Rakudo too?
09:08 mj41 joined
frettled 27416 jani 20 0 321m 261m 8504 R 100 3.3 1:17.38 parrot 09:08
Hah, I remembered correctly regarding memory usage. :D 09:09
This is on an x86_64 computer, though, so memory usage is higher than it would be on x86_32.
It's still climbing, around 400 MiB virtual now.
I recall that we had some problem reports last year about computers running out of memory while running Parrot to make Rakudo, tee-hee. 09:10
And there it broke the 512 MiB resident barrier. 09:11
Warp 1?
09:12 Axius left
jnthn morning o/ 09:13
09:14 masak joined
masak oh hai, zebras! 09:14
jnthn hah, I beat masak to proving my awakeness by 1 minute :P 09:16
frettled Pandas. We're pandas.
jnthn: \o/
masak frettled++ # preserving our image
jnthn You make it seem to black and white...
*so 09:17
frettled munches on some bamboo sprigs.
jnthn considers a bacon sarnie
masak frettled: you realize that you're not a *real* panda until you have the panda badge on modules.perl6.org? :)
frettled masak: uhoh
Perhaps I'll lay off the bamboo sprigs until I know my stomach can handle it, then. D 09:18
masak waking up, I had this idea about attribute initialization and BUILD and all that stuff. 09:19
I haven't thought it through since that half-asleep state, but I'd like to hear what you think about it :)
frettled Is it coherent? :D 09:20
masak think so.
basically, we ditch the restriction of "you can't access $!foo because this is a type object, sonny".
frettled Hmm, building Rakudo Star 2011.01 took less than 15 minutes on less-than-current hardware. I'll be darned yellow, that's less than I recall it was for the 2010 editions. 09:21
masak when we see an assignment 'has $!foo = "OH HAI";', we do that assignment at class construction time, to the type object.
jnthn wait, what?
masak not a good idea? :)
jnthn Well, type objects are "empty instances". 09:22
It's the lack of storage that makes them undefined.
arnsholt frettled: IIRC there have been a number of Parrot performance improvements in the last year or so
masak no, it's being undefined that makes them undefined.
jnthn And how do we decide that it's undefined? 09:23
frettled arnsholt: yes, I recall some of the chatting about it.
jnthn Anyway, doing it at class construction time is hard.
frettled arnsholt: funnily enough, this hardware is just over four years old :)
jnthn We don't have all the attributes yet, and maybe not all the parents. We need to know to compute a storage approach for, e.g. P6opaque. 09:24
And we can't do that until after class composition.
So in that case the type object really *doesn't* get allocated with anywhere to store the attributes.
I'm guessing where you're going is "and then we just clone the type object to make an instance", though :) 09:25
frettled jnthn: I'm just waiting to hear why that is bad as well :) 09:26
jnthn It'd work in the case masak++ suggested
But for has $.c = $.a + $.b;
It's rather more problematic, as you want to do some evaluation there each time. 09:27
frettled Hmm, I thought $! and $. were disjunct. 09:28
jnthn Well, I coulda written has $!c = $!a + $!b; as well :)
09:28 hanekomu joined
frettled (: 09:28
jnthn (Also, has $!a = foo(); has the issue that foo() might to be defined by class construction time...it may be post declared.) 09:29
mberends ncow: US$2.50 should get you 1GB RAM to massively upgrade your computer: us.webuy.com/product.php?sku=SMEMM8DHD 09:30
masak as I said, it's just an idea, and I was curious where it would lead us; if it solves any problems, etc. 09:32
yesterday I was creating a bunch of small classes just for giving names to things.
they looked like this: 'class OpType::Comma is OpType { method tightness { 2 } };' 09:33
my first instinct was to write it as 'class OpType::Comma is OpType { has $.tighness = 2 }'
but I realized I would only be using these classes as type objects, never as instances. so that idea was out. :/ 09:34
frettled Isn't declaring tightness as a constant very restricting?
masak frettled: I was basically after a way to assign a constant to the type.
so that I can later do 'if $op1.tightness < $op2.tightness { ... }' 09:35
jnthn std: class Foo { constant $.bar = 2; }
p6eval std 4608239: OUTPUT«ok 00:01 121m␤»
masak ah, that would be it.
jnthn Doesn't need an attribute, should be able to just have a constant and a method to access it :)
Granted Rakudo almost certainly doesn't yet handle that. 09:36
masak jnthn++
jnthn :(
masak no, rakudo doesn't handle 'constant'. alpha might've.
jnthn Well
I think it "handled" it. :)
masak jnthn: but that's not your fault :) it's everybody's "fault" :)
jnthn Part of what I've been doing in nqp/ctmo is letting us have a much more proper implementation of constant :) 09:37
frettled masak: Ah, I think I see what you're getting at now.
masak \o/
jnthn: right. 09:38
09:42 woosley joined
masak rakudo: class C {}; use Test; is_deeply C.new, C.new, "the are alike" 09:48
p6eval rakudo 5ac05e: OUTPUT«not ok 1 - the are alike␤# got: C.new()␤# expected: C.new()␤»
masak s/the/they/
anything we can do about that? or would that just be overkill?
jnthn Well, the fallback could be "are they exactly the same type, and if so, do they deeply have the same attributes" 09:52
masak that's what I'm thinking.
oh well, I'll just write an adapter this time.
turning the objects into hashes.
jnthn .Capture.hash will pretty much do that for you, btw :) 09:54
masak even for private attributes?
09:55 icwiener joined
jnthn ah, good question 09:55
forget offhand
masak rakudo: class C { has $!a }; say C.new(:a(42)).Capture.hash.perl 09:56
p6eval rakudo 5ac05e: OUTPUT«{}␤»
09:56 zby_home_ joined
masak rakudo: class C { has $.a }; say C.new(:a(42)).Capture.hash.perl 09:56
p6eval rakudo 5ac05e: OUTPUT«{"a" => 42}␤»
masak jnthn++
and, phew. :)
rakudo: class C { has $!a; method a { $!a } }; say C.new(:a(42)).Capture.hash.perl 09:57
p6eval rakudo 5ac05e: OUTPUT«{}␤»
masak hm...
jnthn It'll go by .has_accessor on the meta-attribute. 09:58
masak nod. 09:59
10:04 MayDaniel joined
masak rakudo: given 42 { when SomeUndeclaredTyp { say "OH HAI" }; default { say "OH NOES" } } 10:07
p6eval rakudo 5ac05e: OUTPUT«===SORRY!===␤Missing block at line 22, near "; default "␤»
masak er, s/Typ/Type/
std: given 42 { when SomeUndeclaredTyp { say "OH HAI" }; default { say "OH NOES" } } 10:08
p6eval std 4608239: OUTPUT«===SORRY!===␤Function 'SomeUndeclaredTyp' needs parens to avoid gobbling block at /tmp/ggW4TXS4y1 line 1:␤------> given 42 { when SomeUndeclaredTyp⏏ { say "OH HAI" }; default { say "OH NOE␤Missing block (apparently gobbled by
..'SomeUndeclaredTyp'…
masak I like STD's error messages much better here.
masak submits rakudobug
10:13 MayDaniel left
jnthn Hmm, may not be too hard to add a basic version of that 10:13
masak rakudo: role A { our sub foo { say "OH HAI" } }; A::foo
p6eval rakudo 5ac05e: OUTPUT«Cannot find sub A::foo␤ in main program body at line 1␤»
masak jnthn: should that work?
jnthn mumble 10:14
masak submits rakudobug
jnthn :)
I don't know, tbh.
I can argue it both ways.
masak me too.
but I wrote it by accident, and now that I'm thinking about it, it doesn't seem so odd.
jnthn role A { our sub foo { say $?CLASS } } # ouch :)
masak well, yeah. but there will always be edge cases. 10:15
jnthn True
frettled masak: cutting edge cases, even :)
masak I can distinguish methods using the $obj.A::foo syntax, so why not subs using just A::foo?
frettled: that's why they're double-edged swords? :P 10:16
jnthn rakudo: sub foo($x) { our sub bar() { say $x; } }; bar()
p6eval rakudo 5ac05e: OUTPUT«Null PMC access in type()␤ in 'bar' at line 22:/tmp/5gk1BaNW3F␤ in main program body at line 22:/tmp/5gk1BaNW3F␤»
jnthn It's basically this :)
frettled masak: or used as razors by Mr. Occam? 10:17
jnthn std: sub foo($x) { our sub bar() { say $x; } }; bar()
p6eval std 4608239: OUTPUT«Potential difficulties:␤ &bar is declared but not used at /tmp/GLTX5X17cG line 1:␤------> sub foo($x) { our sub bar⏏() { say $x; } }; bar()␤ok 00:01 122m␤»
jnthn Interesting. :)
masak I know STD is a bit wrong in that area.
so take that with a granum of salus. 10:18
jnthn It complained about &bar not being used, but not about bar() not being defined.
masak this is untenable. the sun is shining, and I haven't had my run this weekend. 10:23
as lue++ would've said: "curse you, nice weather!"
frettled See masak run.
10:24 satyavvd joined 10:26 satyavvd_ joined 10:27 awoodland joined 10:28 satyavvd left 10:34 sftp_ joined, sftp left 10:38 awoodland left 10:50 whiteknight joined 10:52 satyavvd_ left 11:07 dayangkun_ joined 11:12 Chillance joined
dalek rors: 62aa2c5 | moritz++ | notes.pod:
Exception isa Capture
 11:13
masak Capture: Function call arguments (right-hand side of a binding) 11:30
moritz: is this so that you'll be able to match Signatures against Exceptions?
um... we're overloading $0 etc for exceptions? 11:32
that feels... odd.
moritz: I'd have expected the uniquely identifying part of an exception to be the type object of the Exception (a subtype of X:: or whatever). 11:34
moritz: why use strings or integers when we have a type system?
moritz masak: because I don't want to force people to create a new type for every possible error 11:39
masak: note that you can still create enums of errors, and stick them in as IDs
this is not java after all :-) 11:40
masak I'm not saying it's Java.
I am saying, however, that I looked forward to the interaction between smartmatching on X:: types and inheritance. 11:41
I think the long-term benefits of that should be at least considered.
moritz you can create X:: enums, and smartmatch .[0] against them 11:42
you can also subclass yourself, or compose roles into your own exception objects 11:43
they are just objects after all
11:44 icwiener left
masak I'm thinking of the use case "catch any numeric exception" or "catch any I/O exception". 11:49
moritz masak: would you be happier with coarse grained types and carrying finer grades of information via the capture elements? 11:53
11:54 Patterner left 11:56 Psyche^ joined, Psyche^ is now known as Patterner
masak moritz: at this point, it's not so much about happiness. I just want to make sure we don't put a poor man's type system in place where we could have had... our type system. 11:59
I already know 'when' clauses work very well with types. 12:01
12:01 arnsholt left
moritz 'when' clauses work well with everything that smart-matches 12:01
masak yes, but you know what I mean. when I use a type, I have exactly one word, and I write that. 'when SomeType'. 12:02
if you want to talk about happiness, handling exceptions that way would make me happy. :)
moritz I see what you mean, but I have some reservations 12:03
those include: I don't want to define a huge number of empty classes or roles solely for classification. That just feels wrong.
if those types are composed at runtime on demand, that itself can lead to further errors 12:04
masak could you give a concrete example? 12:05
moritz methods can clash when composing multiple roles
or does that only happen at compile time? 12:09
jnthn Depends.
If you mix them in one by one, no clash.
If you do like $foo does (Bar, Baz, WTF)
Then you could get a conflict. 12:10
Note that if matching against signatures is important, that mostly involves defining .Capture to do whatever sort of mapping from the exception object you wish
It doesn't have to be a capture
$foo ~~ :(...) will, as a fallback if none of the other smartmatching cases match, do a capture coercion. 12:11
12:14 dayangkun_ left
masak the "conflict" you get with '$foo does (Bar, Baz, WTF)' isn't an error; it's just later methods applied winning. 12:19
jnthn No
$foo does (Bar, Baz, WTF) is "derive an anonymous subclass, and .^add_role each of them, then compose it" 12:20
(See S14) 12:21
It's different from
(($foo does Bar) does Baz) does WTF
masak ah. 12:24
I sit corrected.
jnthn The semantics you want are available, just not with the syntax you mentioned. :) 12:29
masak right. 12:33
12:34 Chillance left
jnthn stroll & 12:36
12:41 arnsholt joined
masak hunting for food & 12:57
13:05 arnsholt left 13:17 MayDaniel joined 13:27 MayDaniel left 13:32 yegor left 13:35 pmurias joined
jnthn Hunting? Gee, I normally just go to the supermarket these days... 13:40
13:41 alester joined 13:47 yarp joined 14:03 yegor joined 14:08 hudnix joined 14:09 arnsholt joined 14:39 uniejo left 14:41 yarp left
TimToady it seems that std: isn't updating, 'cuz I fixed this several weeks ago 14:46
std: sub foo($x) { our sub bar() { say $x; } }; bar()
p6eval std 4608239: OUTPUT«Potential difficulties:␤ &bar is declared but not used at /tmp/c4_DUQ3F35 line 1:␤------> sub foo($x) { our sub bar⏏() { say $x; } }; bar()␤ok 00:01 122m␤»
14:47 Chillance joined 14:49 yegor left
TimToady and I think roles are the right approach to use on exception classifications 14:49
if you compose at compile time, you'll get conflicts at compile time 14:50
14:50 yegor joined
TimToady and that's exactly what roles are for, inheritance without hierarchy 14:51
we can also ensure that if an error occcurs while creating exceptions, we capture all the data raw and don't thrown any info away 14:52
"Panic: here's everything I know: ..." 14:53
s/ensure/try to ensure/ :)
14:54 MayDaniel joined
TimToady bbl & 14:54
14:55 yegor left 14:59 colomon left 15:01 MayDaniel left 15:04 colomon joined 15:07 pmurias_ joined, pmurias_ left, yegor joined 15:11 kaare_ joined 15:19 sivoais left 15:24 jaffa4 joined 15:38 justatheory joined 15:44 woosley left 15:50 [Coke] left
masak \o/ # roles/inheritance for exceptions 15:51
15:51 [Coke] joined 15:52 daxim left 15:54 hanekomu left 16:13 envi left
masak is there a list equivalent of .join? 16:15
i.e. input a list, output the same list but with one or more items intersprixeled in between each original element.
rakudo: say (1..5 Z "foo", *).perl 16:16
jnthn Zip?
p6eval rakudo 5ac05e: OUTPUT«(1, "foo", 2, "foo", 3, "foo", 4, "foo", 5, "foo")␤»
masak not quite.
I don't want the fencepost at the end.
oh well. 16:17
I'll just write it out as an explanatory for loop :)
frettled TimToady++: oooh, nice clarifying explanation 16:18
16:20 alester left 16:21 icwiener joined
masak for @original { push @new, $_, "foo"; LAST { pop @new } } 16:21
is quite nice.
sorear good * #perl6 16:26
moritz hello sorear 16:27
16:27 hanekomu joined
dalek ecza: 11b04b4 | pmurias++ | hoopl/ (2 files):
[hoopl] refactoring
 16:27
pmurias sorear: hi
jnthn o/ sorear 16:28
jnthn ponders what he's gonna do for dinner :)
moritz just had a lovely, self-made pizza
jnthn ooh :)
frettled is eating two oven-baked fillets of trout with a lemon-mango marinade, plus a mash of cooked vegetables (carrots, sweet potato, etc.). 16:29
jnthn Eww, fish. :P] 16:30
frettled ghoti!
jnthn :D
16:30 hanekomu left
sorear ncow: ping 16:33
masak is getting ready to cook a lentil dal with whole cinnamon, cardamom, and cloves 16:35
perl6: my @foo; pop @foo; say "alive" 16:36
p6eval niecza v4-66-g11b04b4: OUTPUT«Unhandled exception: System.IndexOutOfRangeException: Array index is out of range.␤ at Niecza.VarDeque.Pop () [0x00000] in <filename unknown>:0 ␤ at CORE.C264_List_2Epop_2Ditem (Niecza.Frame ) [0x00000] in <filename unknown>:0 ␤ at Niecza.Kernel.RunCore
..(Niecza.Frame& cur) [0x00…
..pugs, rakudo 5ac05e: OUTPUT«alive␤»
masak sorear: ^
jaffa4 is there pack 16:38
unpack
in Perl 6?
masak jaffa4: yes.
moritz jaffa4: you should know by know where the specs are, and how you can search them
s:2nd/know/now/
jaffa4 In what language it implemented?
masak jaffa4: and I started implementing them last summer, but they're big, so I only got started.
jaffa4: they're in the setting, so they're in Perl 6.
jaffa4 Is it implemented? 16:39
masak jaffa4: partly.
not too much, actually.
patches are welcome.
jaffa4 have your though about convert perl5 functions into perl6?
masak see github.com/rakudo/rakudo/blob/mast...Buf.pm#L40 for the implementation of Buf.unpack 16:40
jaffa4: yes, I have.
jaffa4 that is how you do it
masak jaffa4: specifically, I'd like to port the 17k Perl 5 tests for pack/unpack.
jaffa4 ,is that not?
masak jaffa4: the implementation will likely differ a bit, because Perl 5 doesn't have the concept of Buf.
moritz perl 5 doesn't separate Str and Buf, which makes it non-trivial
masak and that has more repercussions that one might first think. 16:41
moritz also perl 5 doesn't have user-facing scalar data types
masak unpack/pack are full of (conscious) intermixings of string and buf.
moritz so it doesn't matter if unpack returns '123' or 123 in perl 5
it does in perl 6
masak hm, that one never posed that much of a problem.
but yes, it's a difference.
jaffa4 String and buffer? 16:42
I do not know the difference myself?
masak hold on, I have a blog post on that :)
16:42 am0c joined
jaffa4 I do not know the difference myself... 16:42
masak strangelyconsistent.org/blog/str-an...get-it-now
jaffa4: ^
jaffa4 By the way, D language makes similar distinctions 16:44
When does it make a difference? 16:45
Are these going to be two different type?
masak yes, exactly.
strings are made of characters. bufs are made of bytes. 16:46
jaffa4 so string is going to have encoding property I guess
masak bytes are not characters. bytes *code for* characters.
jaffa4: maybe as an implementation detail, they will.
but when something is a string, that's when it doesn't *need* an encoding property. 16:47
because it's already characters.
jaffa4 characters but what characters?
moritz Unicode characters
codepoints or graphemes
masak even that is an implementation detail. though a more important one. 16:48
moritz no, it's really part of the API
masak TimToady seems to not want us to be entirely tied to Unicode, though.
there are some places on the planet where Unicode is a bad idea. 16:49
moritz if I have a base character and a combining character, does . in regex match them both?
masak sounds like a question for tchrist :P
moritz ah, your "even that" was about Unicode, not about graphemes/codepoints, it seems
jaffa4 utf8,16?
32? 16:50
masak moritz: aye.
jaffa4: preferably all of them.
moritz jaffa4: those can all encode the same characters, which is that question doesn't matter
masak jaffa4: that's just a question of how you encode/decode.
jaffa4: but we tend to default to utf8.
jaffa4 ok 16:55
16:59 masak left, masak joined 17:01 pmurias left 17:03 pmurias joined 17:07 knewt2 left
dalek ast: ea701e8 | moritz++ | S02-builtin_data_types/instants-and-durations.t:
[instants-and-durations.t] test Durations with integers to avoid occasional test failures due to FP errors
 17:08
moritz I've meant to do this for ages.
masak moritz++ 17:09
tadzik g'afternoon pandas 17:18
masak tadzik! \o/ 17:19
17:24 am0c left
moritz tadzik++ # I for one welcome our new module overpanda :-) 17:25
tadzik overpanda? :) 17:27
moritz a panda-shaped overlord :-) 17:30
masak .oO( "...things quickly turned panda-shaped" ) 17:31
tadzik I could gain some Panda-weight :0 17:33
:)
thundegnat++! 17:34
A vice-panda appears! Or even the biggest Panda, he has TWO Panda-badged modules :) 17:35
still I don't know why does he use - rather than ::
17:40 mj41 left 17:46 birdwindupbird joined 17:53 birdwindupbird left 18:05 Mowah joined 18:09 _jaldhar left, _jaldhar joined 18:10 SHODAN joined 18:12 Mowah left
tadzik oh, the lovely "Died". What a strange disturbance of the force: tests pass, program dies 18:14
masak time to write more tests! 18:15
(and perhapd not to die without a message from within the application) :)
tadzik "ha ha ha" 18:16
try { $panda.resolve($x) }; 18:17
say $! if $!;
seems to be the problem. Is $! reseted after a successful try {}, and should it?
masak: yeah, the main thing that disturbs me is not the fact that I shipped a broken Panda, but the fact that my tests are wrong :) 18:18
tests do not test panda-the-script after all
18:22 Rotwang joined 18:26 Mowah joined 18:32 [Coke] left 18:34 [Coke] joined 18:41 risou left
tadzik hmm 18:41
try { foo() }; $fail = $! is clearly not the same as try { foo(); CATCH { $fail = $! } }. Intentional? Why? 18:42
18:43 rgrau joined
moritz a try { } that didn't catch an exception resets $! 18:45
so that after the try you can ask $! if something went wrong
tadzik yes 18:46
the first case resulted in a strange "Died" out of nowhere, the second works
rakudo: try { die 'a'; CATCH { die "inside" } } 18:48
moritz that's a know bug
p6eval rakudo 5ac05e: OUTPUT«(timeout)»
tadzik just checking if it's still here
18:59 ponbiki joined 19:01 birdwindupbird joined 19:02 birdwindupbird left 19:03 birdwindupbird joined 19:05 Mowah left 19:06 Alias joined, newguy left 19:08 Alias_ left
tadzik phenny: tell colomon IO::Prompter's META.info has a missing dep (Testing), mind adding it? 19:09
phenny tadzik: I'll pass that on when colomon is around.
tadzik phenny: thank you dear
19:14 Mowah joined
tadzik jnthn | Doesn't emacs have everything? ;) 19:15
lue | It has an organizer! \o/
jnthn | I bet it even has a girlfriend.
we surely need "#perl6quotes", like Parrot has ParrotQuotes on their wiki
jnthn gee, when'd I say that... :P 19:16
19:18 noganex joined 19:19 szr joined, icwiener left
ncow ok, it's been over 40 hours and it's still compiling. 19:20
cognominal I know that text has a mate :)
ncow sweet buttery jesus, what is it in there that could be talking so long? 19:21
tadzik ncow: you're below the system requirements, get over it
ncow actually from what I've read I seem to be right on the line
19:21 noganex_ left
ncow nevertheless, I've never had something take so long to compile on this system. 19:22
arnsholt ncow: It's well established that swap thrashing will instantly increase your run-time by many orders of magnitude
frettled ncow: btw, your hardware is not eight years old, the CPU is at leaste nine years old. :) And you have too little RAM (as I said), so your OS is probably trashing.
arnsholt It sucks, but there not really anything that can be done about it
frettled except upgrading a little bit
ncow: I tested on _four_ year old hardware, hardly anything like current, and got Parrot + Rakudo (from Rakudo Star 2011.01) compiled in about 12 minutes. 19:23
ncow yes the HD activity LED indicator is constantly lit uo
up
may I ask why it needs so much ram just to first-time compile? Again, I feel forced to compare this experience to everything else I've ever compiled from source on this (or other) machines 19:24
frettled ncow: then I'm afraid to say you cannot have compiled much in terms of recent software. That something requires more than 200 MB of RAM while compiling is, unfortunately, hardly news.
sorear ncow: HELLO
ncow why was this process constructed in such a way that calls for so much ram, that is all I want to know. 19:25
sorear ncow: do you have me on ignore or something?
tadzik sorear: no, he just asks the same question for the third time
ncow frettled: I've compiled an awful lot, actually, on this machine alone, as well as countless other set ups, believe me kindly, I'm not stranger to this
sorear: no, sorry
sorear: I'm looking back but I don't see anything from you other than these two lines that I see here, if there is something I've missed, I am sorry 19:26
tadzik: I'm not meaning to sound repetative, it's just a little frusterating. It seems this build process is grossly biased towards newer hardware, when 98% of thing that I've build from source on this machine take under 15 minutes, depending on the size of the package. Sometimes even just a couple minutes. I can build an entire tool chain (glibc, gcc, bin-utils) in under 3 hours 19:28
frettled ncow: even the linux kernel, in less than 15 minutes?!
ncow It can build Perl 5 in under half an hour. Yet it takes almost two days so far to build. 19:29
masak ncow: your computer is swapping a lot.
ncow frettled: as I mentioned the other day, that one is an exception. the kernel image + modules usually takes around 4 hours
frettled ncow: and the problem is that Parrot _does_ require a lot of RAM to do it
ncow 3 hours*
frettled: which is still a far cry from 48 hours
frettled ncow: it's not an exception, I'm afraid, it's rather common for complex software to take both lots of memory while compiling and lots of disk access.
masak ncow: this would all be solved with more RAM.
ncow masak: I have no doubt about that 19:30
frettled: I perfectly udnerstand that
but I still find it difficult to accept that the current process is the most optimal and best all around.
pmurias it's not 19:31
jnthn ncow: Nobody has set out and *planned* for the build process to take a bunch of RAM. It does because the toolchain it's built upon isn't particularly well optimized yet.
frettled ncow: nobody has claimed it is
jnthn Or in some places, at all optimized.
pmurias but your machine is obsolete
ncow When I fly back home, I'll be building it on a newer quad core machine with 8 gigs of ram (my own at home) which i'm sure will be monumentally faster
frettled ncow: in fact, you've gotten explicit acknowledgement that it's suboptimal.
ncow fair enough
frettled ncow: if it's a four year old 2.4 GHz box like the one I tested on (in a multi-user setting), expect around 12 minutes :) 19:32
masak neither of us should be blaming the other. it's just that ncow's setup and Rakudo don't work together well right now.
jnthn ncow: Things will improve, but development takes time. :-)
ncow I'm nto blaming anyone
sorear ncow: have you considered not using Rakudo? 19:33
frettled sorear: heresy! :D
ncow do you fellas suppose I could instead fire up linux in a virtual machine (I have one in vmware ws 7) on my dual core laptop that I have with me here and build it there, and then transfer the finished build to the real box?
19:33 donri left
masak ncow: sure. 19:33
sorear ncow: rakudo builds are non-transportable, that's why there's no binary available for download
ncow I'm not sure what dependacies it relies on. I usually prefer to build on (or at least for) the target system
SHODAN or you could just replace the real box with the laptop 19:34
;)
ncow sorear: I was afraid of that
sorear: not use rakudo? what other Perl 6 alternative is there?
sorear ncow: now, on the other hand, github.com/downloads/sorear/niecza...a-4.01.zip
pmurias niecza uses a lot less ram
ncow sorear: I just wanted to test out Perl 6
sorear I have a 8-9 year old computer with 384 MB ram
tadzik ncow: I understand your mood, and believe me, when you're developing Rakudo, a 15-minute-lasting compilation is not something I fancy a lot. But that's a fact: you're below the memory limit. It's not the length, not the time, it's the amount of memory it needs. What you are probably stuck on, is the compilation of a file of over 7 thousand lines. Rakudo is currently a memory hog, and because it's basically 19:35
compiling itself, it takes a helluva memory. Combined with the fact that you don't really have that much, it may cause funny situations, like a 40-hour-lasting compilation. As I said, this limitation used to give people segfaults (was that you sorear?), so you're partly lucky, and partly not. Now, the process of minimizing Rakudo's memory usage as well as splitting the compilation to a smaller parts is in
progress, ask jnthn for details. We're sorry for the current state of Rakudo's performance, but we really do what we can, given the manpower we have
...whoa, that was a message
ncow: does that clear your doubts and answer your questions?
sorear ncow: Parrot's garbage collector is the butt of many jokes around here
ncow: even the people with 8 GiB lying around think it's pathetic 19:36
ncow tadzik: yes, thanks. I udnerstand all that.
tadzik I'm glad
sorear: oh, it got a lot better recently
jnthn sorear: The generational GC is an improvement at least. 19:37
Not that there isn't room for more improvement but still...it's a move in the right direction.
sorear it still needs to stop requiring 5 separate allocations per P6object PMC 19:38
jnthn Indeed.
sorear it still needs to stop retaining every single temporary in each function (some people call this a problem with the register allocator, but I have a very inclusive conception of "garbage collector") 19:39
ncow: I don't recommend trying to build niecza, it seems to need about 300MB while compiling itself, but the binary I linked you should work fine 19:40
ncow: don't use the Debian mono version (2.6); 2.8 uses about half as much memory (2.10 also works :)) 19:41
19:42 GinoMan joined 19:49 pjcj left 19:57 nymacro left 19:58 alester joined 19:59 justatheory left 20:00 colomon left 20:05 birdwindupbird left, molaf left 20:06 colomon joined 20:09 ymasory left
ncow sorear: thanks 20:12
20:14 birdwindupbird joined 20:27 _jaldhar left, _jaldhar joined, aindilis left 20:28 aindilis joined 20:32 pjcj joined
dalek p/ctmo: 1d1ea2f | jonathan++ | src/metamodel/how/NQP (4 files):
Get packages meta-objects to set a hash in their .WHO (can promote it to some more full-on Stash type later, probably, but this is fine for now.) (Foo.WHO)<$x> := 42 style stuff now works fine.
 20:34
p/ctmo: 06f82a8 | jonathan++ | src/metamodel/knowhow_bootstrapper.c:
Should have a stash for KnowHOWs too.
p/ctmo: 9869549 | jonathan++ | src/NQP/ (2 files):
s/$*PKGMETA/$*PACKAGE/ - it's the compile time $?PACKAGE, after all. Install $?PACKAGE. Set up a GLOBALish, though nothing is done with it just yet.
20:41 tzhs left, wallberg joined 20:49 birdwindupbird left 20:59 SHODAN left 21:00 ymasory joined 21:02 GinoMan left 21:08 GinoMan joined 21:11 Rotwang left 21:13 everyone_loves_P joined
masak everyone_loves_P: \o/ 21:13
everyone_loves_P: I'm more of an M person myself.
moritz
.oO( everbody loves pm? )
 21:14
everyone_loves_P no, Perl!!!!
jnthn oh, truncation :)
everyone_loves_P yeah (((
21:15 everyone_loves_P left
jnthn
.oO( everyone_loves_pivo )
 21:15
21:21 peterhil joined
masak peterhil: hei! 21:24
peterhil Hei 21:32
21:32 kaare_ left
masak peterhil: tervetuloa meidän nöyrä kanava. 21:34
21:41 dorlamm joined 21:42 ymasory left 21:43 GinoMan left
peterhil Kiitos, ajattelin vain vilkaista mitä uutta Perl 6:ssa on... 21:44
Opettelin joskus vähän Perliä ennen Pythonia ja Rubya.
frettled Oooh, I understood five words.
jnthn I understood Perl, Python and Ruby... :) 21:45
peterhil (And now everyone idling will wonder what I'm writing...)
jnthn smiles at the cute case declensions
sorear I think I get the "6" too 21:46
peterhil Translation: Thank you, I just thought to take a peek at what is new on Perl 6... I learned some Perl before Python and Ruby. 21:48
masak peterhil: uskon Perl 6 pitää mukavia yllätyksiä sinulle.
did I get that right? I'm leaning heavily on Google Translate here.
jaffa4 miért kell idegen nyelven beszélni? 21:49
jnthn What was that about a goat-pole?
masak jaffa4: ah, Hungarian. :)
21:50 alester left
jaffa4 sorear: do you have a test suite? 21:50
peterhil masak: Oikeammin: Uskon että, Perl 6 sisältää ...
jaffa4 no comprendo tu palablas 21:51
peterhil So I've been learning Javascript lately and read from Wikipedia that Perl 6 has had influences from it in addition to Haskell (optional strong typing?) 21:52
Are there some functional programming features from Haskell?
Like map, reduce, takeWhile, cycle, foldr etc...?
jaffa4 cycle 21:53
I do not know cycle
peterhil What are the JS influences? The closures and the object system?
jaffa4 typeless?
jnthn peterhil: Yes, Perl 6 has various functional bits
peterhil IIRC cycle takes an array and loops it over in Haskell.
21:54 silent_h_ joined
jnthn rakudo: say [*] 1..10 21:54
p6eval rakudo 5ac05e: OUTPUT«3628800␤»
jnthn Is reduction meta-operator
jaffa4 rakudo [**] 3; 21:55
rakudo: [**] 3;
p6eval rakudo 5ac05e: ( no output )
peterhil jaffa4: Haskell had some special term for the optional typing... I think it was type inference
jaffa4 rakudo: [**] 3,1;
p6eval rakudo 5ac05e: ( no output )
masak rakudo: .say for (1, 2, 3) xx *
jaffa4 rakudo: [**] (3,1);
pmurias peterhil: i don't think Perl 6 was inspired by Javascript much
p6eval rakudo 5ac05e:
..OUTPUT«(timeout)␤3␤1␤2␤3␤1␤2␤3␤1␤2␤3␤1␤2␤3␤1␤2␤3␤1␤2␤3␤1␤2␤3␤1␤2␤3␤1␤2␤3␤1␤2␤3␤1␤2␤3␤1␤2␤3␤1␤2␤3␤1␤2␤3␤1␤2␤3␤1␤2␤3␤1␤2␤3␤1␤2␤3␤1␤2␤3␤1␤2␤3␤1␤2␤3␤1␤2␤3␤1␤2␤3␤1␤2␤3␤1␤2␤3␤1␤2␤3␤1␤2␤3␤1␤2␤3␤1␤2␤3␤1␤2␤3␤1␤2␤3␤1␤2␤3␤1␤2␤3␤1␤2􏿽xE2
rakudo 5ac05e: ( no output )
masak peterhil: there's your cycle. 21:56
and I don't like it that "(timeout)" overwrites the beginning of the output :(
peterhil: and yes, we have map, reduce/fold, and a bunch of other FP concepts. 21:57
peterhil: in fact, Perl 5 (and earlier) already had map and grep.
peterhil masak: OK, cool
pmurias Perl 5 already has closures so they were not taken from javascript 21:58
which itself seem to be a bit Perl inspired
masak Perl 6 has prototypes. sort of.
pmurias masak: they are real prototypes 21:59
masak oh, good.
pmurias masak: it's just that a prototype OO system is left for CPAN
diakopter javascript has $varnames and everything ;)
masak then maybe that's the inspiration from JavaScript?
peterhil In Javascript the objects are basically just hashes. But the values can contain functions. 22:00
masak rakudo: my $obj = { foo => { say "OH HAI" } }; $obj<foo>() 22:01
peterhil The description on Wikipedia for Perl6 seems to describe something similar
p6eval rakudo 5ac05e: OUTPUT«OH HAI␤»
masak peterhil: that's not usually how we do objects, but it definitely works.
rakudo: class C { method foo { say "OH HAI" } }; C.new.foo 22:02
p6eval rakudo 5ac05e: OUTPUT«OH HAI␤»
jnthn sleep & 22:03
pmurias what concepts from python and ruby influenced Perl 6?
peterhil masak: That latter way seems more familiar from some years back, when I started to learn Perl by doing the Useless Python exercises.
At some point I gave up, and continued with learning Python. :-) 22:04
masak I am not familiar with those exercises. 22:05
pmurias: I know Ruby sigils inspired Perl 6 twigils :)
22:05 dual left
masak pmurias: (Ruby uses a '@' sigil to show attribute classitude) 22:05
Python... Python had the chained comparisons before Perl 6 did, I think. 22:06
and though it doesn't have junctions, it has set operations named any() and all() that work a bit similar. 22:07
pmurias remove mention of SMOP from the wikipedia Perl 6 article
* removed
peterhil Mixins from Ruby? 22:10
masak no, from Smalltalk. 22:11
I don't think Ruby's do method collision detection.
oh noes!
rakudo: say [1..3].perl 22:12
p6eval rakudo 5ac05e: OUTPUT«[1, 2, 3]␤»
22:12 alester joined
masak ranges get expanded! 22:12
pmurias rakudo: say [1..1000].perl
p6eval rakudo 5ac05e: OUTPUT«[1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 26, 27, 28, 29, 30, 31, 32, 33, 34, 35, 36, 37, 38, 39, 40, 41, 42, 43, 44, 45, 46, 47, 48, 49, 50, 51, 52, 53, 54, 55, 56, 57, 58, 59, 60, 61, 62, 63, 64, 65, 66, 67, 68, 69, 70, 71, 22:13
..72, 73…
masak rakudo: class C { method postcircumfix:<[ ]>($p) { say $p.WHAT } }; C.new[1..5]
p6eval rakudo 5ac05e: OUTPUT«Range()␤»
masak oh phew.
still in business, then.
pmurias rakudo: say (1..3).perl
p6eval rakudo 5ac05e: OUTPUT«1..3␤»
masak they don't expand in postcircumfixes. :)
pmurias rakudo: say ["a","b",1..3,"d"].perl 22:14
p6eval rakudo 5ac05e: OUTPUT«["a", "b", 1, 2, 3, "d"]␤»
masak I guess circumfix:<[ ]> has some kind of flattening context going on.
the things you learn by writing tests :P
jaffa4 What is []? 22:16
masak rakudo: class O { method postcircumfix:<[ ]>() { 42 } }; say O[] 22:17
p6eval rakudo 5ac05e: OUTPUT«Method '!select' not found for invocant of class ''␤ in main program body at line 22:/tmp/puyVJCCcYF␤»
jaffa4 rakudo: print [1,2,3] 22:18
p6eval rakudo 5ac05e: OUTPUT«1 2 3»
jaffa4 rakudo: print [1,2,3]*3
p6eval rakudo 5ac05e: OUTPUT«9»
jaffa4 rakudo: print (1,2,3)*3 22:19
p6eval rakudo 5ac05e: OUTPUT«9»
jaffa4 rakudo: print (1,2,4)*3
p6eval rakudo 5ac05e: OUTPUT«9»
jaffa4 rakudo: print (1,2,4,5)*3
p6eval rakudo 5ac05e: OUTPUT«12»
peterhil masak: There were some challenges on here back on 2003: replay.waybackmachine.org/200410190...ecode.html
But the excercises I was referring were actually here: www.pythonchallenge.com/ Fun stuff! 22:20
22:20 silent_h_ left 22:21 jaffa4 left
masak hello? p6eval? 22:21
rakudo: say "OH HAI"
it went to bed.
so should I.
'night, #perl6.
22:21 masak left
p6eval rakudo 5ac05e: OUTPUT«OH HAI␤» 22:21
peterhil 'I got to level 20 and did the first six + nr 17 with Perl
'Night masak 22:22
22:26 wallberg left 22:29 coldhead left 22:37 dorlamm left 22:46 dayangkun_ joined 22:47 dayangkun__ joined 22:51 dayangkun_ left 22:57 colomon left 23:03 colomon joined 23:06 rgrau left 23:16 pmurias left 23:19 silent_h_ joined, [Coke] left 23:21 [Coke] joined
diakopter rakudo: say "OH HAI" 23:45
p6eval rakudo 5ac05e: OUTPUT«OH HAI␤»
diakopter notes quickness

Powered by Cro and the Raku® Programming Language.

Please report any issues / comments / feature requests as an issue on App::Raku::Log.

Thank you!